Received: from pub.pub.ro (pub.pub.ro [141.85.128.1]) by postgresql.org (Postfix) with ESMTP id 6FDD7474E4B for ; Sun, 14 Apr 2002 06:48:50 -0400 (EDT) Received: from p16.pub.ro (root@p16.pub.ro [141.85.0.67]) by pub.pub.ro (8.9.3/8.9.3) with ESMTP id OAA22433 for ; Sun, 14 Apr 2002 14:33:19 +0300 Received: from dark (dark.p16.pub.ro [10.2.2.35]) by p16.pub.ro (8.11.6/8.11.6) with ESMTP id g3EAmm323536 for ; Sun, 14 Apr 2002 13:48:49 +0300 Content-Type: text/plain; charset="us-ascii" From: Mihnea Balta To: pgsql-interfaces@postgresql.org Subject: Libpq lock up Date: Sun, 14 Apr 2002 13:56:58 +0300 X-Mailer: KMail [version 1.4] MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200204141356.58031.dark@dark.vodanet.ro> X-RAVMilter-Version: 8.3.0(snapshot 20010925) (compaq) X-Archive-Number: 200204/26 X-Sequence-Number: 2568 Hi. I'm using libpq in several apps. If I keep a connection open for a long time (hours) and pass a large amount of queries through it, it finally locks in a select() syscall. I'm freeing the previous result set (if it exists) before making a new query, so I suppose it's not an out-of-memory/handles problem. My platform used to be linux 2.4.9 / glibc 2.1.3 / postgres 7.2.1. Now I've moved postgres on a linux 2.4.18 / glibc 2.2.3, but in the mean time I modified my apps to buffer stuff and connect to the database each time they need to perform a query, so I don't know if upgrading glibc solved the problem. Has anyone encountered a simmilar problem so far? Now I've got to write an app that performs a fair amount of queries and I'd rather not disconnect/connect to the backend for each one of them. Buffering is not well seen in this case either, because loosing buffered data due to an application crash is bad karma. Thanks in advance ;)